# feed this the source of testall.rnc to get a table of
# content models in desc order of frequency
# LB 5 oct 07

$verbose = 1;

while (<>) {
 if (/element\s+([a-zA-Z0-9]+)\s+\{/) {
  $content = "$1\=";
 } elsif ($content) {
     if (/att\./) {
	 flush()   
      } else {
         $content .=  $_;
      }
   }
}



print "$modelCount models found\n";

foreach $key (sort by_value keys %models) {
	$freq = $models{$key};
	$cumTot += $freq;
	$cumFreq = $cumTot/$modelCount;
	if ($freq > 3) {
             print "$key  \[$freq \] $cumTot ($cumFreq)\n"    ;
	     print "   $modelkey{$key}\n" if ($verbose); 
	} else {
	   $specialModels{$key} ++;
	   $countRare ++;
	}   
  }
print "\nRare models (<3) $countRare \n";
      
foreach $key (sort keys %specialModels) {
	print "$key  $modelkey{$key} \n"; 
}	




sub by_value { $models{$b} <=> $models{$a}; }



sub flush {
my $model; my $gi;
    $content =~ s/\,\s*\n//; 
  $content =~ s/\n//g; 
   $content =~ s/ //g;
   $content =~ /([^=]+)\=\(*(.+)/;
   $model = $2; $gi = $1;
   $modelkey {$model} .= "<$gi> ";
   $models{$model} ++;
   $modelCount ++;
   $content="";
}